Abstract
The report is a collaboration between the Minderoo Centre for Technology and Democracy, the Bennett Institute for Public Policy, and ai@cam which argues that the unregulated use of Generative AI in the UK economy will not necessarily lead to economic growth, and risks damaging the UK’s thriving creative sector.
This report examines the impacts GenAI may have on the creative sector’s workforce and productivity. By exploring the current copyright landscape in the UK and US as it relates to AI, we examine the challenges surrounding licensing agreements, performers’ rights, transparency provisions on AI systems, copyright in AI outputs, and false attribution. Significantly, we consider the Government’s proposed Text Data Mining (TDM) exemption and outline why opt-out frameworks are not fool-proof solutions.
Recommendations
- Holistically examine the impact GenAI is having on the workforce in the creative industries, including by commissioning research on AI adoption across the sector, and use it to inform robust policies for supporting the sector’s workforce.
- Encourage the uptake of licensing agreements to ensure copyright holders are compensated for use of their work by AI systems, but it should also ensure that these licensing agreements fully acknowledge the rights of copyright holders and fairly compensate them for the use of their works.
- Independently ratify and adopt the Beijing Treaty on Audiovisual Performances as a first step in ensuring greater protections on performers’ rights and from false attribution by AI systems.
- Adopt transparency requirements on the training of AI systems which include the mandatory disclosure of data provenance.
- Clarify that only a human author will be afforded copyright in the outputs generated by AI models and produce guidance on: a) the threshold for ‘creative intellectual effort’ in achieving copyright in AI outputs; b) the need for recognition and compensation to artists whose name and canon are used in prompts to AI models that generate outputs; and c) measures required to avoid false attribution in AI outputs.
- We urge caution against embarking on the path of a Text and Data Mining (TDM) exemption, regardless of an ‘opt-out’ mechanism, without a robust economic analysis of the impact that it will have on the creative industries.
Collaborative Approach
This policy brief represents a collaboration between three University of Cambridge groups:
- Bennett Institute for Public Policy
- Minderoo Centre for Technology and Democracy
- ai@cam
Download the Full Report
The complete findings can be found in the full report.